Aaron Howard and Dr. Bret Mackay, Economics
Health Care costs have increased substantially over the past several years. Unfortunately, the cause of the increase is not well understood. The purpose of my research was to study and model the increasing premium trends of HMOs in an attempt to identify factors that contribute to the increasing costs of medical care.
The project was divided into three sections. First, I studied the industry and learned the factors that might contribute to increases in cost. Second, I compiled the statistical data available from the Mercer annual health plan survey. Third, I analyzed the data, looking for factors that are statistically significant.
While studying the industry, it was apparent that the increased cost of HMOs could be quantified using the HMOs’ overall premium trend. The possible factors or variables that might contribute to the increased premium trend are listed below with a short description of each.
Contributing Factors
First, the actuarial accuracy of HMOs’ estimated future medical cost could contribute to increased future premium trend. To measure the accuracy, the medical loss ratio (MLR), which is the percentage of the HMOs’ total premiums spent on medical claims, was studied. I anticipated that as the (MLR) increases, the premium trend will also increase. However, it is possible that the MLR is not linear in its relationship to the future premium trend. An extremely low MLR may indicate that the HMO is overestimating claims to provide a buffer. Overall, it is expected that as MLR declines, the premium trend will decrease at a declining rate. The effect, however, may not be a good indication of overall health care increases because the HMOs would be expected to eventually correct the estimating problem and be more conservative in their estimates. This effect could cause an actuarial cycle. While this cycle may have contributed to the l! atest increases, it is unlikely the problem is strictly faulty actuarial estimations. Thus, the MLR may be a good indicator of the year-to-year increase of medical costs, while failing to accurately model the general increasing costs.
A second variable of interest is a measure of utilization. To estimate utilization, the total number of days HMO members were in the hospital (“bed days”) per thousand members was examined. It is expected that as bed days per thousand increases, the premium trend would also increase, signaling higher overall utilization and contributing to the increased cost.
The ability of the HMO to administer the plan efficiently may also be a contributing factor. To study efficiency, the administrative loss ratio (ALR), the percentage of the premium dollars spent to administer the plan, was used. If a plan is extremely inefficient and has a high ALR, then we may assume that plan’s premium trend will increase to cover the cost. However, an extremely low ALR could be an indicator of poor medical management, which increases utilization. Thus, the relationship of the ALR to premium trend may not be linear, and it may be both positive and negative in different ranges of the ALR.
The plans’ operating profit margin may also contribute to increased premium trends. As profit increases, we would expect the premium trend to decrease. However, if profit grows relatively large, it could be a sign that the HMO is consistently overpricing the product and artificially increasing trends. Thus, the relationship between profit and premium trend may not be linear. Another possible contributing factor examined was NCQA accreditation. NCQA accreditation is recognized as a sign of quality among HMOs. For most consumer products, we would expect to pay some fee to enjoy a higher quality product (or a perceived higher quality product). However, it could be that poor quality HMOs are poor because they lose money and need to increase premiums to cover losses. In short, it is not clear if the NCQA accreditation’s effect on premium trend is positive or negative.
The last two variables studied were total number of members and years of operation. It is not clear if the size of the group and the years of operation have positive or negative effects on the premium trend.
Compiling the Data
Gathering the data for the project was complicated. Mercer Human Resource Consulting sends an annual survey to every HMO in the U.S. Unfortunately, the data was not complete and often inaccurate. This quality of the data required many hours of interpreting the raw data to compile a file that was accurate enough to analyze. Regrettably, many assumptions were necessary to complete the compilation. For example, in the survey, the plans were asked to supply the MLR, ALR, and profit ratio. These statistics must sum to 1, but many of the plans answered incorrectly. To complete the file, it was assumed the MLR and AML were correct, and the profit margin was calculated following the prescribed formula. Plans that did not complete the necessary parts of the survey were omitted. This was particularly troubling because the degrees of freedom were reduced, making significant results more difficult to obtain. Unfortunately, due to incomplete data, t! he overall sample size was only 620. In addition, because many of the values measured should be very precise and many of the plans supplied estimates of the data, the variables had little variation.
Results
The data seemed to indicate that the total membership in the HMO was very significant. As membership increases, the premium trend deceased slightly. This result could help to explain the merger activity observed in the industry. The number of years an HMO has been in operation was the closest other variable to having a significant effect on HMO premiums. This variable was also negative in its relationship to trend.
MLR, AML, and profit were not significant in the analysis. Interestingly, NCQA accreditation was correlated with decreases in premium; however, it was not significant. Perhaps the most interesting variable was bed days per thousand, which was not significant. It is generally believed that increased utilization is the cause of increased cost. While this may be true, this project was not able to validate the claim.
The project was interesting and valuable, although further research is needed in the area to complete the study. In particular, it is important to identify a measure of utilization that could be more accurately used in an analysis. Additionally, it would be interesting to investigate how the compensation increases of medical personnel affects the premium trend.